Sine-wave windowed synthesis

Abstract

Speech can be understood even when a 3-tone replica of speech is presented to a listener. In tonal synthesis, called sine-wave synthesis (SWS), the output signal consists of number of time-varying sinusoids that follow center frequencies and amplitudes of the first number of formants of a natural utterance. In this paper we propose an alternative technique of speech synthesis. It is based on the number of dominant frequency components present in the original signal. In the proposed method, called sine-wave windowed synthesis (SWWS), the amplitudes and frequencies of tonal components are changed in discrete steps in subsequent time windows. Perceptual tests performed on Polish speech show that signals synthesized with the SWWS technique are judged as more natural and intelligible than SWS speech.